Bridging the Micro and Macro Calibration of Agent-Based Model Using Mean-Field Dynamics
ABSTARCT :
Calibration of agent-based models (ABM) is an essential stage when they are applied to reproduce the actual behaviors of distributed systems. Unlike traditional methods that suffer from the repeated trial and error and slow convergence of iteration, this article proposes a new ABM calibration approach by establishing a link between agent microbehavioral parameters and systemic macro-observations. With the assumption that the agent behavior can be formulated as a high-order Markovian process, the new approach starts with a search for an optimal transfer probability through a macrostate transfer equation. Then, each agent's microparameter values are computed using mean-field approximation, where his complex dependencies with others are approximated by an expected aggregate state. To compress the agent state space, principal component analysis is also introduced to avoid high dimensions of the macrostate transfer equation. The proposed method is validated in two scenarios: 1) population evolution and 2) urban travel demand analysis. Experimental results demonstrate that compared with the machine-learning surrogate and evolutionary optimization, our method can achieve higher accuracies with much lower computational complexities.
EXISTING SYSTEM :
? There are different variants of CSS and to take a pluralistic approach to it may be considered wise. CSS could be seen today as a larger umbrella under which different approaches might coexist and somehow feel legitimate.
? On the other hand, when agent models are not derived from any pre-existing agent theory or vision, whether computational or not, but only by the behaviour they are expected to generate (Epstein, 2006), agent models are arbitrary, poorly comparable, competent in highly specific domains of knowledge and disarmingly inapt in any other.
? It would but result either in tight but essentially useless theories, or in accurate predictions of poorly understood social phenomena.
DISADVANTAGE :
? This allows ABMs’ researchers to the hard problem of standard estimation methods in order to offers a true real and feasible alternative to currents economical models.
? In general the problem of determining the parameter values of a system from input-output data is called the estimation problem.
? The identifiability problem is more circumscribed: given a model of the system and specific input-output experiments we ask if the parameters of the model can be uniquely determined.
? The identifiability problem is not specific to econometrics but spreads from statistics to control and system engineering, from chemical to biological problems.
PROPOSED SYSTEM :
• An integrated variant, which takes after ABM, reconciling it with the quantitative one, is proposed as a fundamental requirement for a new program of the CSS.
• This is not equal to finding poorly realistic the model of agent often proposed by current modellers, and asking for a seemingly human one.
• Another important distinction concerns the way in which mental representations are incorporated: symbolic representations allow an agent to mentally manipulate them in order to reason, plan, take decision, communicate, etc.
• The practice of agent based modelling however did represent a substantial underexploitation of such wide spectrum of possibilities.
ADVANTAGE :
? More efficient numerical methods and very powerful large computer allows us to test identifiability in any linear or linearized DSGE models.
? Agent based modeling is a tool used to overcome the limitations of pure mathematical analysis, it allows the construction of more realistic models.
? The main interest of ABMs is that they are very versatiles and can provide extremely valuable tools for generating scenarios, that can be used to test the effect of policy decisions in times of large economic instabilities.
? It is quite clear that an important assumption to obtain convergence is that the statistics used identify the parameters of interest, that is there is one to one relationship between the theoretical values of the statistics and the value of the parameters.
|